Counting Available Parking Space Using Image Processing Information Technology Essay
Nowadays, peoples are facing problem to find an available parking space in parking lot due to the tremendous increase of occupancy of cars. When driver enters a certain parking lot, the driver takes a long time just to find an available parking space. A Counting Available Parking Space using Image Processing (CAPSuIP) will develop to solve the problem that driver faced with low cost. CAPSuIP will use image processing to detect of existence of the car and also provide information such as number of available parking space and the location of that parking. The system will capture image using webcam and process the image to counting available parking space. The system use a modified Software Development Life Cycle (SDLC) to plan, analyze, design, development and testing. The development of this system will use techniques of image processing that will implement in each phase of methodology. This system will give information about the location of available parking space and the number of available parking space. It will be benefit to all drivers when enter a parking lot.
1. Introduction,
Nowadays, car is very important to everyone especially for who are works. People are willing to make installment to get an own car. When talking about metropolitan, then traffic jam always occur because of numbers of vehicles are so high. Thus we cannot deny the existing of the cars in our daily life. Whenever we go out by car, we are facing problem to find an available parking space due to the tremendous increase of occupancy of cars.
The analogy is when driver enters a certain parking lot, the first thing that the driver do is looking forward of some sign to telling that the parking lot is fully occupied, partly occupied or vacant. The driver also do not know how many are there and where to find a parking division for his/her car. Some of parking divisions may remain unoccupied even the total occupancy is high. This will causing ineffective use of parking divisions as well as traffics jams around the entrance of parking lot. Therefore, by offering drivers with relevant information on the parking lot during entering a parking lot becomes an important issue.
The proposed system called as Counting Available Parking Space using Image Processing (CAPSuIP). This system proposes a method of detecting the existence of parked vehicles by processing the image of the parking lot taken by a surveillance camera and then counting the available parking space which is display in front of entrance of parking lot.
The system employ images, since all area in the parking lot can be observed with relatively few camera. Other than that, the system is compact and the cost is not is not expensive. The image of a parking lot is taken by a surveillance camera set at some height in the parking lot.
1.1 Problem Statement
There are some reasons why Counting Available Parking Space using Image Processing (CAPSuIP) is developed. The problems that have been identified are stated below:
Driver needs some relevant information before entering the parking lot such as the current available parking spaces in the parking lot.
There are current system used in parking lot but the method used is based on the detection by installing a certain sensor on each division; the other is to detect cars through images of the parking lot taken by surveillance cameras. In the method with the sensor, the cost rises as the number of parking divisions because a lot of sensors are required corresponding to each parking divisions.
1.2. Objectives
Objectives of the Counting Available Parking Space using Image Processing (CAPSuIP) to be developed are to:
Capture and detect existence of vehicle at parking lot using image processing technique.
Count, display of the available parking spaces in parking lot.
1.3. Scope
There are a few scopes that have been identified in order to develop the system. The scopes of the systems are:
This system is just a prototype system using image processing techniques.
Using image that captured from webcam.
The position of the parked vehicle is correct.
The location of case study of the system is at Universiti Malaysia Pahang (UMP) parking lot block Z. Each location consists of five space of parking block.
The system can be used in daytime only without have a strong shadow.
2. Literature Review
2.2.1 Comparison between Car-Park Occupancy Information System (COINS) and Locating Vehicle in a Parking Lot by Image Processing
A Car-Park Occupancy Information System (COINS) [1] is developed to be a viable solution to reduce the amount of time needed to search for a vacant car-park lot especially in a huge parking area. With this system, images captured by a surveillance camera were processed in real-time to identify the occupancies of the parking lots. This occupancy information is further processed by a central control unit and distributed to display panels located at strategic locations at the parking area. The drivers can easily find a vacant parking lot based on the information displayed on the panels. Motivation for developing this system came from the fact that minimum cost is involved because image processing technique is used rather than sensor-based techniques. As surveillance cameras are readily available in most car parks, this technique is much cost effective than installing sensor on each parking lot.
Locating Vehicle in a Parking Lot by Image Processing[2] is more concern to propose a method of detecting the existence of parked vehicles by processing the image of a parking lot taken by surveillance camera. Whenever driver wants to park a car at a parking lot, how to find a proper parking division there causes a serious problem. The objective of the present article is in providing drivers with such information as the lot is fully occupied or relatively vacant, where unoccupied parking divisions are found, and so on. The images employed, since all areas in the parking lot can be observed with relatively few cameras, the system is compact, and the cost is not expensive. The image of a parking lot is taken by a surveillance camera set at some height in the parking lot.
The relevant issues are how to cope with both temporal and spatial changes in illumination, how to discriminate shadows from vehicles, how to cope with occlusion, and how to cope with various surface reflectances of vehicles and so on. To cope with these issues, the input images transformed to the gray levels with log-transform, extracts edges and counts the number in each parking division, and then decides if each division is occupied or not. The recognition rates for a set of images taken at various moments of a day were well above 95 %.
2.2.2 Comparison between Parking Guidance System using RFID and Image Processing Techniques in WSN Environment and Parking Space Vacancy Monitoring
Parking Guidance System using RFID and Image Processing Techniques in WSN Environment [3] describes a novel approach to developing a Parking Guidance System within the car park in a Wireless Sensor Network (WSN) environment in order to help alleviate the frustration and problem in finding vacant parking space. The system utilizes the existing CCTVs installed in the car park coupled with FPGA device in detecting the vacant spaces which will in turn be assigned to the patron using the shortest path algorithm based on both the point of entrance to the car park and building. The patron is then guided to the specified location by referring to the map printed on the parking ticket. Besides that, an RFID tag is also attached to the parking ticket to uniquely identify the assigned parking space of the patrons and will be used to remind patrons of the parking location during payment.
Whereas, Parking Space Vacancy Monitoring[4] is propose a stereo-vision based system that deal with instances with severe vehicular occlusion. In this system, multiple cameras are used to monitor the vacancy status of the P502 parking spaces on University of California, San Diego (UCSD) campus.
In this system, a method for monitoring vacancies in parking lots using a stereo camera system presented to create a 3D reconstruction of the scene, which enables us to determine the vacancy status of a particular parking space under vehicular occlusion. Additionally, results for 3D reconstruction using uncalibrated versus calibrated cameras are compared.
This system is able to identify vacancies while differentiating between spaces for different permit holders (faculty versus students). Ideally, the system also able to provide an exact count of the number of available spaces, but it must have to appeal to a statistical notion of vacancy, as certain spots may be too heavily occluded by trees, other vehicles, etc. to be monitored with very high accuracy. This information will ultimately be integrated with a status dissemination tool, where drivers will be able to query the parking lot status via mobile phone.
3. Methodology
The figure 3.1 shows the flow of the Software Development Life Cycle (SDLC) methodology that has been modified in developing Counting Available Parking Space using Image Processing techniques.
Figure 3.1: Software Development Life Cycle.
3.1 Analysis
Requirements and previous system information analysis defined in the literature review chapter that include in chapter 2 of the thesis. It is include existing system information to analyze the technique used and the implementation of the system.
Data are gather from the image acquisition process where there are a lot of images had been captured.
3.2 Design
This phase will describe about the process flow involved in developing the system. There eight step which is the image acquisition, initiation processing, image feature extraction, grayscale subtraction, image edge detection, RGB color subtraction and so on. Figure 3.2 shows the image processing flow for counting available parking space.
Figure 3.2: Process flow in CAPSuIP.
3.2.1 Image Acquisition
Image acquisition is a first stage of any vision in image processing. After the image has been obtained, various method of preprocessing can be applied to perform the many different vision tasks. The locations of image to capture are in the UMP parking area. The scopes of the system are only five (5) space of parking block in the UMP parking area block Z. Other than that, the system using model simulation to analyze the image.
These images will capture by the author using the webcam 5megapixels. Image that will used is 480 x 640 pixel using JPEG format. JPEG format is use because JPEG images are best used for the representation of natural scene.
3.2.2 Image Feature Extraction
In this phase, image will extracted to identify the location of every parking lot in the image. There are two types of images are used which is image without a car and image with a car.
An image without a car will processed to identify the coordinate of parking lot. The height and width of the parking lot will be identifying based on the line in parking lot. After the height identified, it will be divided into five (5) to separate the each location of parking lot. Figure 3.3 shows of the illustration of the parking lot which is separate into five blocks.
Figure 3.3 Illustration of block of the parking lot
An image with a car processed to identify the existing of the car in the parking lot.
3.2.3 Initiation Processing
Preprocessing is second phase in the phase of Counting Available Parking Space using Image Processing. Preprocessing phase is a process of improvement of digital image without knowledge about the source of degradation. It can used to improve an image’s contract and brightness characteristics, reduce its noise content, or sharpen its details. There are two (2) steps in the image enhancement to improve the quality of an image which is converting image into double and normalize RGB image.
Converting Image Into Double
The images are used in this system will be converted to double type of images. This process will converts the true color image RGB to double precision, rescaling the data if necessary. This conversion is important because it will make the calculation more accurate.
Normalization RGB Image
Image that capture will be processed by normalize the RGB image to remove the effect of any change in intensity. This process is very important phase before move to next process.
3.2.4 Removal Shadow
To get an object which free of shadow, three (3) steps must be done. First step is RGB color subtraction, second step is thresholding the image and last step is dilating the image.
RGB Color Subtraction
Image without car and image with car will be subtracted to get the different value between images. This process will execute after the image is normalize.
Thresholding Image
Thresholding process is to get shadow free image.
Dilating Image
Image will be dilated to structuring element object, or array of structuring element objects. This process will be used after the process of subtraction of the images.
3.2.5 Foreground Classification
Exact boundary of the objects and the shadows of the object are needed. To do so, below steps are applied:
Gray Level Transformation
Gray level transformation is used to convert RGB image to gray scale.
Grayscale Subtraction
The grayscale image without car and with a car will be applied.
Thresholding Image
Thresholding process is to get shadow free image.
Dilating Image
Image will be dilated to structuring element object, or array of structuring element objects. This process will be used after the process of subtraction of the images.
Filling Image
Image will be filled out the holes in a bounded area of an image.
3.2.6 Reconstruction and Object Classification
To get the reconstructed boundary of the object without shadow region, point wise multiplication of objects and it is shadow is multiplied with the dilated object region. Figure 3.4 shows the example code of reconstruction of the object.
finalRes(:,:,1) = Res(:,:,1).*re;
finalRes(:,:,2) = Res(:,:,2).*re;
finalRes(:,:,3) = Res(:,:,3).*re;
Figure 3.4 Example of code of reconstruction
To find the area of object, code below is used:
[B,L] = bwboundaries(final);
[Z,N]=bwlabel(final);
if N > 0
for m=1:N,
[r,c]=find(Z==m);
area=size(c,1);
if (area>saiz1 && u1>5)
condition1 = 1;
end
end
end
N = number of label connected component
area = size of selected label
saiz1 = size of image
u1 = number of edge detected
condition1 is determine whether car exist or not. If condition1 is 1 then the block have a car.
To find minimum value of images by grayscale it and use code below:
t1=min(min(fil));
fil = the grayscale image of [2]
t1 = minimum value of selected image
If minimum value greater than 20, then proceed to RGB color subtraction + grayscale.
Subtract RGB color image [2] and then grayscale it.
If size detected image is greater than 40% and less than 90% then car is exist in that block.
4. Implementation
In this phase, there are several process that must be done before getting the result. The input that needed are image without a car as a background and image with a car as a foreground. The example of input image is shown in Table 4.1.
Table 4.1 Example of input
Background
Foreground
5.1 Data Reduction
In data reduction phase, first step is the image was segment from original image into specific location of parking using their width and height and axis X and Y. Second step is the segment image from image that segmented before into 5 block of location of parking by dividing the width of the image. The sources code for the implementation is same as above.
5.2 Conversion Of Image Into Double Type
Conversion image into double type format is actually converts the true color image RGB to double precision, rescaling the data if necessary. This conversion is important because it will make the calculation more accurate.
5.3 Normalization Of RGB Image
Method of converting an RGB image into normalized RGB is used to removes the effect of any intensity variations. After aggregating R, G and B, it is checked that if any value is 0 then it is set 0.001 to overcome “divide by zero errorâ€Â. Next process is to normalizing R factor and G factor.
5.4 Subtraction Between Background Image And Image With Car
Subtraction method is used to get new image which taken by subtract background image and image with car.
5.5 Thresholding The Image
Thresholding method is applied to get shadow free image.
5.6 Dilating The Image
Dilating image is applied to structuring element object, or array of structuring element objects.
5.7 Gray Level Transformation
Gray scale transformation is a process of transformation from RGB image to the gray scale image using gray scale transformation.
5.8 Subtraction Of Gray Image Between Background Image And Image With Car
Subtraction of gray image method is also used to get new image which taken by subtract background image and image with car. After that, the image will be converted to double type.
4.9 Thresholding The Gray Image
Thresholding method is applied in gray image also to get shadow free image.
4.10 Dilating And Filling The Image
Dilating is applied to structuring element object, or array of structuring element objects and filling method is used to fill out the holes in a bounded area of an image.
4.11 Reconstruction The Image
Reconstruction image is used to reconstructed boundary of the object without shadow region, point wise multiplication of objects and it is shadow is multiplied with the dilated object region.
4.12 Regrayscale And Closing The Image
Regrayscale is used to convert rgb image to gray image and closing method is used to performs morphological closing on the grayscale or binary image IM, returning the closed image,. The structuring element must be a single structuring element object, as opposed to an array of objects.
4.13 Feature analysis
Results of feature analysis phase were defined after feature extraction phase where the classification of image will define and count the available parking space. Technique that uses to get the extraction value of image is size and edge of object detected in the image. The value of object fix to be greater than 15% and the edge fix to be greater than 10. If size and edge is fulfill the criteria, then the system will decide that there is a car.
If not, minimum value of image will be determine. Minimum value fix to be greater than 20 and if that criteria is fulfilled, then it will proceed to the RGB color subtraction + grayscale process. If size detected image is greater than 40% and less than 90% then car is exist in that block.
After that, rectangle will be display in the result to show that the block of parking is having a car. The result will appear after all process has done.
4.14 Interface Counting Available Parking Space using Image Processing Technique
The interface of the system is shown in the Figure 4.1. This system can capture image directly from webcam or browse image in the local computer.
Figure 4.1 Interface of Counting Available Parking Space
5. Result and Discussion
5.1 Result Analysis
To test the system, there are two method can be used namely test in real environment and using simulation model. The position of webcam that used must be placed in suitable height and parallel with parking lot. The system that developed has been developed is can capture directly from webcam and also browse image in local computer.
But to test using this method is difficult because every changes of the parking lot must be captured manually. So, the real time system has been developed to testing the system easily. The real time system will detect every changes of the parking lot, the system automatically detect and shows result without interaction from user.
Area to be processed
– This is the background image which is used as
input of the system.
– Parking space fixed to five (5) space only
– This is the foreground image with red line.
– Red line will appear after all process detection
object done.
– The system detect 4 cars in the parking lot. It is
means that only one (1) parking space is
available.
Webcam
Parking Block
Car Model
USB Webcam
5.2 Assumption and Further Research
These parts were explained about assumption that face by author when finishing the thesis and an idea for a further research.
5.2.1 Contraints
In order to complete this prototype, there are two (2) constraints that affect the smoothness of the implementation. There are technical knowledge and lack of an experience.
Technical Knowledge
This project is using MATLAB (Matrix Laboratory) software and ADOBE PHOTOSHOP. It will take time to learn the software and familiar with it.
The System Itself
The system can used only daytime and if there is any strong shadow, it cannot detect whether it car or not. Other than that, the position of webcam and background must be fixed. In addition, the system not well tested because of not enough of needed equipment such as high quality webcam which is used to test the system in real world environment.
6. Conclusion
For the further research, it is recommended to improve the methodology, technique and algorithm to get the better result and can solve the strong shadow problem.
There are a lot of technique that had been applied nowadays and other developer can apply the prototype by using other technique and application. Besides that, the developer can apply the prototype with new algorithm to produce the better result. Other than that, other developer can develop real time system so that it can be used in real world environmnet.
This project can be applied in many places such as university because it will help user to know how many available of parking in a short time.
Order Now